phd.dibris.unige.itphd.dibris.unige.it/biorob/media/PhD themes 33o... · Envisioned control...
Transcript of phd.dibris.unige.itphd.dibris.unige.it/biorob/media/PhD themes 33o... · Envisioned control...
PhD Program in Bioengineering and Robotics
Curriculum Humanoids and Advanced Robotics
Research themes
1. NONLINEAR CONTROL OF HUMANOID ROBOTS FOR BIPEDAL LOCOMOTION ................................. 4
2. DEEP LEARNING FOR LEARNING OBJECT AFFORDANCES ............................................................ 5
3. ROBOT DESIGN FOR DEXTEROUS MANIPULATION ................................................................... 6
4. AUTONOMOUS LEARNING OF OBJECTS USING MULTIMODAL, EVENT DRIVEN CUES .......................... 7
5. IMPROVED GRASPING THROUGH ED VISION AND AFFORDANCES ................................................ 8
6. COMPRESSIVE SENSING AND LOW-LATENCY EMBEDDED PROCESSING FOR THE ICUB ...................... 9
7. MULTIMODAL PERCEPTION OF OBJECTS ............................................................................. 10
8. MULTIMODAL OBJECT EXPLORATION AND GRASPING ............................................................ 11
9. SENSING HUMANS: ENHANCING SOCIAL ABILITIES OF THE ICUB PLATFORM ................................. 12
10. SCENE ANALYSIS USING DEEP-LEARNING ............................................................................ 13
11. TOWARDS FLYING HUMANOID ROBOTS ............................................................................ 14
12. SIMULTANEOUS MULTIMODAL FORCE AND MOTION ESTIMATION ............................................. 15
13. CONTROL OF HYBRID FLOATING BASE SYSTEMS FOR HUMANOID ROBOT LOCOMOTION ................ 16
14. ROBUST STATE ESTIMATION FOR HUMANOID ROBOTS OPERATING ON NON-RIGID CONTACTS ....... 17
15. MOTION STRATEGIES FOR MULTI-LEGGED ROBOTS IN UNSTRUCTURED ENVIRONMENTS ................. 18
16. ROBOT-ASSISTED MICROSURGERY ................................................................................... 19
17. NOVEL INTERFACES AND TECHNOLOGIES FOR ASSISTIVE ROBOTIC SYSTEMS ................................ 20
18. LEGGED ROBOTS (HUMANOIDS) FOR HAZARDOUS ENVIRONMENT INTERVENTIONS ..................... 21
19. LEGGED ROBOTS (QUADRUPEDS) FOR HAZARDOUS ENVIRONMENT INTERVENTIONS .................... 22
20. ADVANCED TELEOPERATION: TELE-LOCOMOTION, TELE-MANIPULATION, AND HUMAN-MACHINE INTERACTION FOR
EXTREME ENVIRONMENTS. ................................................................................................... 23
21. INTUITIVE CONTROL OF INDUSTRIAL EXOSKELETONS .............................................................. 24
22. ADVANCED PHYSICAL HUMAN-ROBOT INTERFACES FOR EXOSKELETONS .................................... 25
23. LEGGED LOCOMOTION-CONTROL: LOCOMOTION CONTROL OF A MOBILE MANIPULATION PLATFORM WITH
HYBRID LEG AND WHEEL FUNCTIONALITY .................................................................................. 26
24. LEGGED LOCOMOTION-CONTROL: WHOLE-BODY SENSOR FUSION AND MOTION ADAPTATION OF A LEGGED
MANIPULATION ROBOT ....................................................................................................... 27
25. LEGGED LOCOMOTION-CONTROL: ROBUST LOCOMOTION OF HUMANOIDS IN AN UNSTRUCTURED ENVIRONMENT
UNDER EXTERNAL FORCES .................................................................................................... 28
26. LEGGED LOCOMOTION-CONTROL: DEXTEROUS HUMANOID WALKING ON RESTRICTED AND UNSTABLE
FOOTHOLDS ...................................................................................................................... 29
27. CONTROL: HUMAN-ROBOT COLLABORATIVE CONTROL OF A MOBILE MANIPULATION PLATFORM WITH HYBRID LEG
AND WHEEL FUNCTIONALITY .................................................................................................. 30
28. CONTROL: WHOLE-BODY CONTROL OF HYBRID HUMANOID-QUADRUPED ROBOTIC PLATFORM ... 31
29. ROBOTIC ACTUATION: NEW EFFICIENT ACTUATION SYSTEMS BASED ON THE BLENDING OF FORCED AND NATURAL
DYNAMICS ........................................................................................................................ 32
30. ROBOTIC ACTUATION: DESIGN PRINCIPLES AND CONTROL OF HIGH PERFORMANCE ROBOTIC ACTUATION SYSTEMS
33
31. ROBOTIC ACTUATION: DESIGN AND CONTROL OF NOVEL LIGHTWEIGHT ROBOTIC JOINT-LINK MODULES WITH
DISTRIBUTED VARIABLE STIFFNESS ........................................................................................... 34
32. MULTIMODAL PERCEPTION: 3D PERCEPTION FOR ROUGH TERRAIN LOCOMOTION AND FREE-FORM OBJECT
MANIPULATION ................................................................................................................. 35
33. MULTIMODAL PERCEPTION: CUTANEOUS AND KINESTHETIC SENSING FOR ROBOTIC ARMS, DEXTROUS HANDS AND
FEET 36
34. WEARABLE ROBOTICS: DEVELOPMENT OF UNDER-ACTUATED UPPER LIMP WEARABLE SYSTEMS FOR TELEOPERATION
AND REHABILITATION .......................................................................................................... 37
35. WEARABLE ROBOTICS: DEVELOPMENT OF EXO-MUSCLES: WEARABLE ADD-ON ELASTIC POWER FORCE
AUGMENTATION UNITS ........................................................................................................ 38
36. LOCOMOTION PLANNING AND ADAPTATION STRATEGIES FOR MULTI-LEGGED ROBOTIC PLATFORMS ON SOFT TERRAINS
39
37. FORCE HARVESTING WITH SHORT-TERM CONTACTS FOR MULTI-LEGGED ROBOT LOCOMOTION ON MOVABLE AND
SLIPPERY TERRAIN .............................................................................................................. 40
38. FLEXIBLE SYSTEMS DEVELOPMENT FOR INDUSTRIAL APPLICATIONS .......................................... 41
39. INDUSTRIAL ROBOTICS FOR INSPECTION AND MAINTENANCE .................................................. 43
In the spirit of the doctoral School on Bioengineering and Robotics, the goal of the advanced and humanoid robotics curriculum is to study the design, realization, programming and control of anthropomorphic and legged robots. Students will work at the forefront of
mechatronics and computer science research jointly covering the full development cycle from
software to mechanical design and from machine learning to realization of sensors, actuators
and electronics. We address the development of the technologies for the next generation of
robots for sensing, actuation and computation. The goal is to develop robots that can
adaptively interact with their environment, learn from their mistakes, and succeed in
performing safely and reliably in real-world environments. Foreseen applications for
anthropomorphic robots range from real-world practical scenarios -e.g., at home, as personal
assistants- to industry as co-workers, to natural or man-made disaster scenarios. Humanoid
robot software deals with vision, audition and tactile perception as well as the ability to look,
reach and manipulate the world while walking freely to reach their targets, interacting
naturally with the environment and their human teachers . The PhD themes in this curriculum are offered by the iCub Facility, by the Department of
Advanced Robotics (ADVR) at the Genova Headquarters of the Istituto Italiano di Tecnologia
(IIT) and by the Center for Micro-BioRobotics (CMBR), in Pontedera (Pisa), part of the IIT
multidisciplinary research network. The iCub Facility is the main integrator of IIT’s research and technology on the iCub humanoid robotic platform. The iCub is the humanoid robot child designed to support researchers
interested in the themes of learning, control, cognition, and interaction, both at IIT and
worldwide. The goal of the iCub Facility is to lead the development of the iCub, arrange and
time the construction of new versions, supervise the incorporation of new technologies and
possibly foster their commercial exploitation. We create opportunities for collaboration at IIT
and worldwide in a large network of iCub owners via European funded projects or
commercial contracts. The iCub Facility ideal candidates are students with a master’s degree in engineering, computer science, physics or related disciplines, open to learning, to novelty
but keeping always an eye on the possibility of implementing research on the state of the art
iCub humanoid robot.
Research within the ADVR concentrates on an innovative, multidisciplinary approach to
humanoid design and control, and the development of novel robotic components and
technologies. This encompasses activities from both the hard and soft systems areas of
robotics. In particular, research on humanoid robotics at ADVR mostly focuses on the COMAN
humanoid robot. The development of the COMAN body exploits the use of actuation systems
with passive compliance, with two main goals: i) to reduce the distinction between plant and
controller that is typical in traditional control engineering to fully exploit complex body
properties, and ii) to simplify perception, control and learning and to explore how compliance
can be exploited for safer human robot interaction, reduced energy consumption, simplified
control, and faster and more aggressive learning. Moreover the last years, there is a particular
attention to the Transfer Technology with a special lab that aims to make the ADVR robots
suitable for industrial applications.
International applications are encouraged and will receive logistic support with visa
issues, relocation, etc.
1. Nonlinear Control of Humanoid Robots for Bipedal Locomotion
Tutor: Daniele Pucci, Ugo Pattacini, Giorgio Metta
Department: iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/research/lines/icub
Description: The general purpose of providing humanoid robots with some degree of locomotion has
driven most of research in the humanoid robotics community of the last decade. Legged and wheeled
locomotion, for instance, have proven to be feasible on various humanoid robot platforms, which can now
be envisioned as interfaces for user assistance in several domains. The intrinsic humanoid robot
underactuation, however, combined with the (usually large) number of the robot degrees of freedom
partially account for the continued attention the robotics community is paying to the locomotion problem.
Although the performances of the state-of-the-art humanoid robots are promising, we are still a far cry
from providing humanoid robots with robust locomotion capabilities.
The research project aims at designing control strategies for providing humanoid robots with bipedal
locomotion capabilities. Walking on soft terrains, partial footholds, and rough terrains are among the
envisioned scenarios. The main challenges for these cases are: i) the modelling of the pair robot-
environment and the identification of the governing system dynamics; ii) the design of controllers with
proven stability and robustness proprieties; iii) the development of robust, efficient implementations of
the control algorithms. Envisioned control solutions make use of Lyapunov and feedback linearisation
techniques [1,2].
The control algorithms will be implemented and tested on the state of the art iCub humanoid robot, one of
the few humanoid robots fully torque controlled (http://y2u.be/VrPBSSQEr3A).
Requirements: the candidate needs to have an engineering or mathematical background with strong
competences in control theory and computer science.
References:
Khalil K. Hassan, Nonlinear Systems, third edition, 2001.
Isidori A., Nonlinear Control Systems, 1995
Contacts: [email protected], [email protected], [email protected]
2. Deep learning for learning object affordances
Tutors: Vadim Tikhanoff, Giorgio Metta
Department: iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/research/lines/icub
Description: In recent years, deep learning methods allowed progress in the deployment of machine
learning to many difficult problems, which in turn dramatically increased the performance of applications
such as computer vision, speech recognition, product recommendation, and a large amount of problems
where large amounts of labeled data can be easily obtained. However, this revolution has had a relatively
small impact in robotics, as a result of the specificity of low-level sensory data and the non-negligible
amount of time it requires to collect them from a robot. Some studies have applied deep learning methods
trained with robot data, showing promising results on a number of tasks, such as reaching, grasping or
manipulation, but the ongoing applications are still quite narrow and few in number.
The concept of affordances can provide a promising framework to study the interaction of a robot with its
environment, and thus to predict the consequences of its own actions on certain objects or with certain
tools.
We are seeking for a highly motivated candidate interested in studying learning of affordances from large
quantities of visuo-motor data. The aim of the project is to enable the iCub to learn autonomously the
effects of its own actions, by gathering the data required to train the network with minimal human
supervision. This knowledge will be applied for action and or/object selection to achieve specific tasks.
Requirements: the candidate is required to have a computer science background with strong competences
in mathematics and machine learning. Proficiency in a programming language is required.
References:
Zhang F, Leitner J, Upcroft B and Corke P (2016a) Vision based reaching using modular deep networks: from
simulation to the real world.
Levine, S., Pastor, P., Krizhevsky, A., Quillen, D. Learning Hand-Eye Coordination for Robotic Grasping with
Deep Learning and Large-Scale Data Collection. ISER 2016
Mar, T., Tikhanoff, V., Metta, G., Natale, L. "Self-supervised learning of grasp dependent tool affordances
on the iCub Humanoid robot." Robotics and Automation (ICRA), 2015 IEEE International Conference on.
IEEE, 2015.
Contacts: [email protected], [email protected]
3. Robot design for dexterous manipulation
Tutors: Alberto Parmiggiani, Giorgio Metta
Department: iCub Facility: https://www.iit.it/research/lines/icub
Description: The unique capabilities of the human hand and arm enable us to perform extremely fine and
dexterous movements, such as those needed to write or to juggle. The emergence of these capabilities was
undoubtedly essential in human evolution: our physical characteristics yield a dexterity that is unparalleled
in the animal kingdom. Moreover, our manual skills are an important part of our interaction with the
environment and of our capacities for feeling, exploring, acting, planning, and learning.
Researchers have been trying to replicate the capabilities of natural systems in artificial ones, by designing
and building robotic hands and arms. Numerous progresses have been made in the past 30 years, but the
dexterity of robotic hands is still far from their biological counterparts. Indeed, further developments in this
area are necessary and can suggest insights regarding the processes underlying dexterous manipulation.
For instance, creating better tactile sensing allows us to observe the touch patterns accompanying human
hand actions. Designing different anthropomorphic hand and wrist shapes provides us with insights into the
role of geometry for hand dexterity. Enabling robots to cope with a multitude of objects and actions
challenges our understanding of how such skills need to be represented and coordinated.
This project will be carried out on the hand-forearm assembly of the iCub robot. In the initial part of the
project, we will explore cutting edge technologies and experiment methods to enhance the hand and wrist
capabilities. We will then work to integrate and implement the most promising approaches on the robot
platform, and ensure they are sufficiently dependable to serve as a basis for advanced and dexterous
manipulation tasks.
Requirements: Successful candidates should be highly self-motivated to pursue independent and
interdisciplinary research. A background in Mechanical Engineering or related disciplines is required.
Hands-on experience with analysis, design and implementation of robotic systems, as well as sensors and
actuators, is strongly preferred.
References:
Johannes Lemburg and Frank Kirchner, Co eptual and embodiment design of robotic p otot pes Int. J.
Human. Robot. 08, 419, 2011 (DOI: http://dx.doi.org/10.1142/S0219843611002526)
Ashish D. Deshpande, Zhe Xu, Michael J. Vande Weghe, Jonathan Ko, Lillian Y. Chang, Benjamin H. Brown,
David D. Wilkinson, Sean M. Bidic, and Yoky Matsuoka, Me ha is s of the Anatomically Correct Testbed
(ACT) Ha d . IEEE/ASME Trans. on Mechatronics, 2013.
Perla Maiolino, Marco Maggiali, Giorgio Cannata, Giorgio Metta and Lorenzo Natale, A flexible and robust
large scale capacitive tactile system for o ots . IEEE Sensors Journal, 2013.
Contacts: [email protected], [email protected]
4. Autonomous learning of objects using multimodal, event driven cues
Tutors: Chiara Bartolozzi, Lorenzo Natale
Department: iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/research/lines/icub
https://www.iit.it/research/lines/humanoid-sensing-and-perception
Description: To effectively interact with the environment and adapt to different contexts and goals, robots
need to be able to autonomously explore and learn about objects. To this aim, we need machine learning
strategies that allow to plan exploratory actions and take advantage of information from multiple sensory
modalities. In particular, we will consider learning from haptic (touch, force and proprioception), auditory
and visual cues (extracted from event based as well as conventional frame based cameras), obtained during
exploratory actions, to investigate how different features contribute to object discrimination.
In the first part of the project we will implement behaviours that allow the robot to interact with objects
through manipulation using predefined explorative procedures (like touching, power grasp, squeezing and
contour following). Events in any sensory channel will trigger acquisition of features from all the available
sensors. In the second part of the project, these features will be used to train machine learning algorithms
for object recognition. The objective of this project is also to investigate to what extent features from
different sensory channels contribute to object discrimination.
The outcome of this project will be a set of explorative procedures allowing a humanoid robot to interact
with objects and extract features from multiple sensory channels and signal processing algorithms for
detecting relevant events during object exploration that trigger feature extraction and identifying a set of
features that allow to discriminate objects using multiple sensory modalities.
Requirements: degree in Computer Science or Engineering (or equivalent) and background in Computer
Vision and/or Machine Learning. High motivation to work on a robotic platform and good programming
skills.
References:
Benosman, R.; Clercq, C.; Lagorce, X.; Sio-Hoi Ieng; Bartolozzi, C., "Event-Based Visual Flow," Neural
Networks and Learning Systems, IEEE Transactions on, vol.25, no.2, pp.407,417, Feb. 2014, doi:
10.1109/TNNLS.2013.2273537
Ciliberto, C., Smeraldi, F., Natale, L., Metta, G., Online Multiple Instance Learning Applied to Hand Detection
in a Humanoid Robot, IEEE/RSJ International Conference on Intelligent Robots and Systems, San Francisco,
California, September 25-30, 2011.
Ciliberto, C., Fanello, S.R., Santoro, M., Natale, L., Metta, G. and Rosasco, L. "On the impact of learning
hierarchical representations for visual recognition in robotics." In Intelligent Robots and Systems (IROS),
2013 IEEE/RSJ International Conference on, pp. 3759-3764.
Contacts: [email protected] [email protected]
5. Improved grasping through ED vision and affordances
Tutor: Chiara Bartolozzi, Vadim Tikhanoff, Ugo Pattacini
Department: iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/research/lines/icub
https://www.iit.it/research/lines/humanoid-sensing-and-perception
Description: The interaction of robots with the environment and the successful completion of service tasks
strongly depend on its capability of grasping and manipulating objects in a robust and reliable way. The
performance of a robot when interacting with objects heavily relies on robot calibration, 3D scene
reconstruction and accurate action execution. In realistic environments all of the above is prone to errors
that can result in failure of the planned action. In humans, such errors are counterbalanced by adaptive
behaviours that build on the knowledge of the results of actions and on the early detection of failures that
allows for the execution of corrective actions. The goal of this project is to exploit machine learning to
devise the dynamics/behaviours of objects during clumsy or imprecise manipulation and grasping, and
using such learned behaviours to plan and perform corrective actions. A bad grasping will cause the object
to fall in specific ways, the robot can then learn the association between a specific action parameter (for
example hand pre- shaping, direction of grasping, object 3D configuration, etc.) and the trajectory of the
falli g o je t, effe ti el lea i g its falli g affo da es . The o ot a the use this i fo atio to ette plan the next grasping action on the same object and to perform corrective actions.
The fine temporal resolution of ED vision will be used to track objects and end effectors of the robot, for
implementing fast closed-loop control of grasping and manipulation. Proprioceptive, tactile and visual
feedback before, during and after grasping will be used to learn the effect of different types of grasping and
actions of the robot on the objects. This method will improve the smoothness of the action, by the
observation of the result of clumsy manipulation, correcting unsuccessful movements, and will allow to
perform online corrective actions to stabilise and re- grasp unstable objects.
This research line will deliver a method for robust, smooth and flawless object grasping by means of fast
sensory feedback and reaction, and improved movement planning
Requirements: degree in Computer Science or Engineering (or equivalent) and background in Computer
Vision and/or Machine Learning. High motivation to work on a robotic platform and good programming
skills.
References:
Glover, A., Bartolozzi, C. (2016, October). Event-driven ball detection and gaze fixation in clutter.
In Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International Conference on (pp. 2203-2208). IEEE.
Pattacini, U. et al., 2010. An Experimental Evaluation of a Novel Minimum-Jerk Cartesian Controller for
Humanoid Robots. Taipei, Taiwan, s.n., pp. 1668-1674.
Mar, T., Tikhanoff, V., Metta, G., and Natale, L., Self-supervised learning of grasp dependent tool
affordances on the iCub Humanoid robot, in Proc. IEEE International Conference on Robotics and
Automation, Seattle, Washington, 2015
Contacts: [email protected] [email protected] [email protected]
6. Compressive Sensing and Low-Latency Embedded Processing for the iCub
Tutor: Chiara Bartolozzi
Department: iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/research/lines/icub
Description: Full (and long lasting) energetic autonomy is crucial for the deployment of robots in daily life.
Efficient sensory encoding is a way to reduce data acquisition, transmission, storage and processing,
leading to power optimisation. Traditional sensing is based on the clocked-driven sampling of data from
se so s e.g. i ages f o a e as, p essu e alues f o at i es of ta els i the ski , et … , leadi g to the acquisition of an enormous quantity of redundant data when the sensory input is unchanged over time
(e.g. images from a camera in front of a static scene), and to large latency due to the acquisition of a full
a a of se si g ele e ts. The iologi all i spi ed e e t-d i e se so e odi g sa ples the sti ulus when the stimulus itself changes of a given amount. Only sensing elements that detect a change trigger
communication and processing, with extremely short latency and high temporal resolution (1us).
The iCub robot features event-driven vision sensors, as well as FPGA-based electronics to compress clocked
readout of tactile sensors exploiting event-driven encoding. The sensors are continuously sampled, but
communication wakes-up at stimulation, with a mean reduction of the data-transfer up to 20% of the
clocked-based encoding, sensibly improving transmission latency, bandwidth and power. The same
infrastructure will be used to efficiently compress other sensors of the robot (e.g. encoders,
accelerometers, etc..). At the same time, the processing power of FPGAs will be used to further compress
the signal by real-time embedded pre-processing (e.g. for filtering, feature extraction, optical flow, saliency,
etc.) and to generate fast reactive control.
This research line will hence deliver efficient sensory encoding and embedded processing for the
development of autonomous robots, encompassing event-driven encoding of multiple sensory modalities
and embedded computation for low latency real-time perception and control that will integrate with higher
level (and slower) processing on the robot.
Requirements: degree in Computer Science, Electronic Engineering (or equivalent) and background in
Computer Vision or Signal Processing. High motivation to work on a robotic platform and good
programming skills also on FPGA.
References:
Bartolozzi, C., Natale, L., No i, F. a d Metta, G., Ro ots ith a se se of tou h , Nat Mate , ol. 9 , pp.921–925, 2016, doi:10.1038/nmat4731
Benosman, R., Clercq, C., Lagorce, X., Sio-Hoi Ieng, Bartolozzi, C., "Event-Based Visual Flow," Neural
Networks and Learning Systems, IEEE Transactions on, vol.25, no.2, pp.407,417, Feb. 2014, doi:
10.1109/TNNLS.2013.2273537
Glover, A., Bartolozzi, C. (2016, October). Event-driven ball detection and gaze fixation in clutter.
In Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International Conference on (pp. 2203-2208). IEEE.
Contacts: [email protected]
7. Multimodal perception of objects
Tutor: Lorenzo Natale, Lorenzo Rosasco
Department: Humanoid Sensing and Perception Laboratory, iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/lines/humanoid-sensing-and-perception
Description:
Conventionally, robots rely on vision to perceive and identify objects. Although computer vision has
recently made remarkable progress, touch and, in general, haptic information can still provide
complementary information. This is because some material and object properties are simply not accessible,
difficult to be estimated from vision (like the object weight, roughness), or even hidden by occlusions.
During active object manipulation multi-modal information is available to the robot and can be used for
learning. Yet, during recognition only partial information may be available (typically vision). This project will
investigate how to integrate visual and haptic information for object discrimination. We will initial consider
the case in which multi-modal information is available during training and recognition. In a second stage of
the project we will investigate how learning with multi-modal cues can help object discrimination when
only partial information is available (vision or touch).
This project will be carried out on the iCub robot. The robot sensory system includes cameras for vision,
tactile sensors, position sensors and force/torque sensors. In the initial part of the project we will build a
dataset for the experiments, by acquiring multi-modal data while the robot grasps a set of objects in
various ways. We will then investigate methods for feature extraction (in particular, for vision, we will
consider features from Deep Convolutional Neural Networks) and machine learning methods (e.g. multi-
view learning, learning with privileged information) for object discrimination using individual and combined
features.
Requirements: the ideal candidate would have a degree in Computer Science, Engineering or related
disciplines; a background in control theory and machine learning. He would also be highly motivated to
work on robotic platform and have computer programming skills.
References:
Higy, B., Ciliberto, C., Rosasco, L., and Natale, L., Combining Sensory Modalities and Exploratory Procedures
to Improve Haptic Object Recognition in Robotics, in IEEE-RAS International Conference on Humanoid
Robots, Cancun, Mexico, 2016
Pasquale, G., Ciliberto, C., Rosasco, L., and Natale, L., Object Identification from Few Examples by Improving
the Invariance of a Deep Convolutional Neural Network, in Proc. IEEE/RSJ International Conference on
Intelligent Robots and Systems, Daejeon, Korea, 2016, pp. 4904-4911
Contacts: [email protected], [email protected]
8. Multimodal object exploration and grasping
Tutor: Lorenzo Natale, Ugo Pattacini
Department: Humanoid Sensing and Perception Laboratory, iCub Facility, Istituto Italiano di Tecnologia
https://www.iit.it/lines/humanoid-sensing-and-perception
Description:
Object shape is fundamental for planning grasping. Precise models of objects may not be available due to
lack of object models, noise in the sensory system or simply occlusions. Object models can be however
inferred by actively explore objects and extracting information from the sensory system. In this project we
will investigate techniques for object modelling and object exploration using multi-modal cues. The main
idea is that vision can provide initial guesses to guide the exploration and that this guess can be
consequently refined using tactile information.
This project will develop novel techniques for automatic object modelling using multi-modal sensory
information. The goal of this project is to advance the state-of-the-art in the field of automatic object
modelling and, consequently, grasping and manipulation of unknown objects. This project will be carried
out on the iCub robot using the stereo system on the robot and the tactile sensors in the hand. We will use
depth information about the objects to build an initial estimation of the shape of the object (e.g. using
superquadrics), we will then implement an active strategy that based on this initial guess will provide the
robot with a sequence of points to be explored with the tactile system. This information will be then used
to improve the model of the object and provide an accurate shape. Validation will be carried out on
grasping tasks.
Requirements: the ideal candidate would have a degree in Computer Science, Engineering or related
disciplines; a background in control theory and machine learning. He would also be highly motivated to
work on robotic platform and have computer programming skills.
References:
Jamali, N., Ciliberto, C., Rosasco, L., and Natale, L., Active Perception: Building Objects' Models Using Tactile
Exploration, in IEEE-RAS International Conference on Humanoid Robots, Cancun, Mexico, 2016
Vezzani, G., Pattacini, G., Natale, L. A Grasping Approach Based on Superquadric Models, in IEEE
International Conference on Robotics and Automation, 2017
Björkman, M., Bekiroglu, Y., Högman, V., and Kragic, D., Enhancing visual perception of shape through
tactile glances, in 2013 IEEE/RSJ International Conference on Intelligent Robots and Systems, 2013, pp.
3180–3186.
Contact: [email protected], [email protected]
9. Sensing humans: enhancing social abilities of the iCub platform
Tutor: Lorenzo Natale, Alessio Del Bue
Department: Humanoid Sensing and Perception lab (iCub Facility) and Visual Geometry and Modelling
lab (PAVIS department)
Description: there is general consensus that robots in the future will work in close interaction with humans.
This requires that robots are endowed with the ability to detect humans and interact with them.
However, treating humans as simple animated entities is not enough: meaningful human-robot interaction
entails the ability to interpret social cues. The aim of this project is to endow the iCub with a fundamental
layer of capabilities for detecting humans, their posture and social behaviour. Examples could be the ability
to detect if a person is attempting to interact with the robot and to react accordingly. This requires a new
set of computational tools based on Computer Vision and Machine Learning to detect people at close
distance. This "face-to-face" scenario requires developing novel algorithms for coping with situations in
which large areas of the body are occluded or only partially visible.
Requirements: This PhD project will be carried out within the Humanoid Sensing and Perception lab (iCub
Facility) and Visual Geometry and Modelling Lab (PAVIS department). The ideal candidate should have a
degree in Computer Science or Engineering (or equivalent) and background in Computer Vision and/or
Machine Learning. He should also be highly motivated to work on a robotic platform and have strong
computer programming skills.
Contacts: [email protected], [email protected]
10. Scene analysis using deep-learning
Tutors: Lorenzo Natale, Lorenzo Rosasco
Description: machine learning, and in particular deep learning methods, have been applied with
remarkable success to solve visual problems like pedestrian detection, object retrieval, recognition and
segmentation. One of the difficulties with these techniques is that training requires a large amount of data
and it is not straightforward to adopt them when training samples are acquired online and autonomously
by a robot. One solution is to adopt pre-trained convolutional neural networks (DCNN) for image
representation and use simpler classifiers, either in batch or incrementally. Following this approach DCNNs
have been integrated in the iCub visual system leading to a remarkable increase of object recognition
performance. However, scene analysis in realistic settings is still challenging due to scale, light variability
and clutter. The goal of this project is to further investigate and improve the iCub recognition and visual
segmentation capabilities. To this aim we will investigate techniques for pixel-base semantic segmentation
using DCNNs and object detection mixing top-down and bottom-up cues for image segmentation.
Requirements: This PhD project will be carried out within the Humanoid Sensing and Perception lab (iCub
Facility) and Laboratory for Computational and Statistical Learning. The ideal candidate should have a
degree in Computer Science or Engineering (or equivalent) and background in Machine Learning, Robotics
and possibly in Computer Vision. He should also be highly motivated to work on a robotic platform and
have strong computer programming skills.
References:
Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, Andrej
Karpathy, Aditya Khosla, Michael Bernstein, Alexander C. Berg and Li Fei-Fei, ImageNet Large Scale Visual
Recognition Challenge, arXiv:1409.0575, 2014
Jon Long, Evan Shelhamer, Trevor Darrell, Fully Convolutional Networks for Semantic Segmentation, CVPR
2015
Pasquale, G., Ciliberto, C., Odone, F., Rosasco, L., and Natale, L., Teaching iCub to recognize objects using
deep Convolutional Neural Networks, in Proc. 4th Workshop on Machine Learning for Interactive Systems,
2015
Contacts: [email protected], [email protected]
11. Towards Flying Humanoid Robots Tutor: Daniele Pucci, Francesco Nori, Giorgio Metta
iCub Facility: https://www.iit.it/research/lines/icub
Dynamic Interaction: https://www.iit.it/research/lines/dynamic-interaction-control
Description: Nonlinear control techniques for humanoid and flying robots have developed along different
directions, and suffer from specific limitations. Besides the morphological differences between aerial and
humanoid robots, one of the reasons accounting for this divergence is that humanoid robot control is often
addressed assuming the robot attached to ground. In this case, the robot is referred to as fixed-base. The
limitations of this approach arise when attempting to tackle the general locomotion control problem, which
also requires the robot to make and break contacts to achieve locomotion. At the modelling level, the
Euler-Poincare` formalism provides one with singularity free equations of motion for the humanoid robot.
In this case, the robot is referred to as floating base: this formalism may allow the development of control
techniques for the general locomotion problem, thus unifying flying and humanoid robot control
techniques.
As a matter of fact, we believe that there is a strong technological benefit in conceiving robotic platforms
capable of contact locomotion, flight, and manipulation. In this respect, flight and manipulation have
already been implemented on many platforms, thus contributing to the so-called aerial manipulation [1],
but robots combining the three aforementioned capacities are still missing.
This research project aims at developing control methods for flying humanoid robots, thus contributing
towards the development of a unified control approach for flying and humanoid robots. The main
challenges for this project are:
i) dealing with the robot under actuation;
ii) dealing with estimation of robot states;
iii) implementation of efficient control algorithms.
The research project will be also focusing on the choice of the humanoid robot propulsion and on
preliminary tests on a real robotic platforms.
Requirements: the candidate needs to have an engineering or mathematical background with strong
competences in control theory and computer science. Competences in robotics and optimization will be
positively evaluated.
Reference:
[1] K. Kondak, et al., U a ed aerial systems physically interacting with the environment: Load
transportation, deployment, and aerial a ipulatio , in Handbook of Unmanned Aerial Vehicles. Springer,
2015, pp. 2755–2785.
[2] Pucci, D.; Traversaro, S.; Nori, F. "Momentum Control of an Underactuated Flying Humanoid Robot".
Available online at: https://arxiv.org/pdf/1702.06075.pdf
Contacts: [email protected], [email protected], [email protected]
12. Simultaneous multimodal force and motion estimation Tutor: Francesco Nori, Daniele Pucci
Department: Robotics, Brain and Cognitive Sciences (Istituto Italiano di Tecnologia)
http://www.iit.it/rbcs
Description: human motion capture (MoCap) finds applications ranging from entertainment (movies and
games) to rehabilitation and sport activity monitoring. Standard marker-based technologies (e.g. the Vicon
http://www.vicon.com) have been recently replaced in several applications with innovative marker-less
systems such as the Xsens MVN (http://www.xsens.com/products/xsens-mvn) and the Microsoft Kinect
(http://www.xbox.com/kinect). The main limitation of these systems is that they provide only kinematic
measurements, typically position, velocities and accelerations. Starting from a technology previously
developed and tested on the iCub [1], this research proposal aims at developing an innovative wearable
technology for human force and motion capturing [2]. This information will be used for reactive and/or
predictive control of movements during physical human-robot interaction. Applications are foreseen in the
field of exoskeletons (with Matteo Laffranchi [3] and Jesus Ortiz [4]) and collaborative manipulation (with
Arash Ajoudani [5]). This research activity is conducted within the European project An.Dy
(http://www.andy-project.eu).
Requirements: the candidate needs to have an engineering background with strong competences in
computer science and basic statistics. Competences in robotics and control will be positively evaluated.
Reference:
[1] M Fumagalli, S Ivaldi, M Randazzo, L Natale, G Metta, G Sandini, F Nori. Force feedback exploiting tactile
and proximal force/torque sensing. Autonomous Robots 33 (4), 381-398
[2] Latella, C.; Kuppuswamy, N.; Romano, F.; Traversaro, S.; Nori, F. Whole-Body Human Inverse Dynamics
with Distributed Micro-Accelerometers, Gyros and Force Sensing. Sensors 2016, 16, 727.
[3] https://www.iit.it/research/lines/rehab-technologies-inail-iit-lab
[4] Toxiri, S., Ortiz, J., Masood, J., Fernández, J., Mateos, L. A., & Caldwell, D. G. (2015, December). A
wearable device for reducing spinal loads during lifting tasks: Biomechanics and design concepts.
In Robotics and Biomimetics (ROBIO 2015).
[5] L. Peternel, N. Tsagarakis, and A. Ajoudani, "Towards Multi-Modal Intention Interfaces for Human-Robot
Co-Manipulation", IEEE International Conference on Intelligent Robots and Systems (IROS), 2016.
Contact: [email protected], [email protected]
13. Control of Hybrid Floating Base Systems for Humanoid Robot Locomotion Tutor: Francesco Romano, Daniele Pucci, Francesco Nori
iCub Facility: https://www.iit.it/research/lines/icub
Dynamic Interaction: https://www.iit.it/research/lines/dynamic-interaction-control
Description: current state of the art walking algorithms for humanoid robots are based on simplified
representations of the robot dynamics, which seldom consider the interactions between the robot and its
surrounding environment. Usually, the robot trajectories are generated by exploiting stability criteria such
as ZMP or Capture Point, and these trajectories are then stabilized by inverse kinematics based controllers.
However, to fully exploit the dynamic capabilities of state of the art humanoid robots, taking into account
both the full system dynamics and the robot-environment interaction is of paramount importance.
This research project aims at developing control methods for floating base systems subject to jumps in their
configuration space, namely, hybrid floating base systems. The control framework is then particularly useful
for humanoid robots since they can be modeled as hybrid floating base systems during locomotion on rigid
contacts. Simulation and control of hybrid systems poses various difficulties. For example, switching events
must be properly detected, thus impacting the underlying governing system dynamics. Event-based ODEs
are the most diffuse tool to accomplish this task, but they are sensitive to Zeno conditions. The challenge of
this project is to advance current state-of-the-art nonlinear control methods for their application on: (1)
hybrid systems, (2) whole-body formulation and (3) real-time systems.
The developed control algorithm will be implemented and tested on the state of the art iCub humanoid
robot, one of the few humanoid robots fully torque controlled (http://y2u.be/VrPBSSQEr3A). It will render
the robot capable of moving in a real-world environment, e.g. walking at a sustained pace, with the
possibility to traverse uneven and/or inclined terrains or climb stairs.
Requirements: the candidate needs to have an engineering or mathematical background with strong
competences in control theory and computer science. Competences in robotics and optimization will be
positively evaluated.
Reference:
[1] Khalil K. Hassan, Nonlinear Systems, third edition, 2001.
[2] Lygeros et al., Hybrid Systems: Modeling, Analysis and Control.
http://inst.cs.berkeley.edu/~ee291e/sp09/handouts/book.pdf
Contact: [email protected], [email protected], [email protected]
14. Robust State Estimation for Humanoid Robots operating on Non-Rigid
Contacts Tutor: Silvio Traversaro, Daniele Pucci, Francesco Nori
Department: Dynamic Interaction Control Laboratory - iCub Facility (Istituto Italiano di Tecnologia)
https://www.iit.it/research/lines/dynamic-interaction-control
Description: humanoid robots need reliable estimates of their state (position, velocity, acceleration,
internal joint torques and external forces) for locomotion and manipulation. Current estimation techniques
either ignore the contacts made by the robot, or assume that this contacts are rigid and bilateral. For
achieving operational autonomy in unstructured environments , humanoids of the future will need to
estimate their state exploiting (rather then neglecting) the presence of non-rigid contacts.
This research project aims to identify models of non-rigid contacts suitable to use in realtime state
estimation. The resulting models will be then be fused with the information coming from the rich set of
sensors available on the iCub robot (6-Axis Force/Torque sensors, Robotic Skin, Distributed Inertial Sensors,
RGB-D sensors) to provide a reliable state estimation to be used by the high level control algorithms. The
main challenge is to extend existing estimation techniques to work on: (1) soft contacts (such as soft soil,
sand or carpets) and (2) contact involving articulated structures (like a door or a seesaw).
Emphasis will be placed on the robustness of the resulting estimation techniques, for enabling use of the
iCub on real world scenarios, e.g. walking on carpets and manipulating pillows in an actual apartment.
Requirements: the candidate needs to have an engineering, computer science or mathematical
background. Prior experiences in robotics, probability theory, tribology and/or C++ programming will be
positively evaluated.
Reference:
[1] C Ciliberto, L Fiorio, M Maggiali, L Natale, L Rosasco, G Metta, F Nori. Exploiting global force torque
measurements for local compliance estimation in tactile arrays. IEEE/RSJ International Conference on
Robots and Systems, 2014.
[2] F Nori, N Kuppuswamy, S Traversaro. Simultaneous state and dynamics estimation in articulated
structures. IEEE/RSJ International Conference on Robots and Systems, 2015
Contact: [email protected], [email protected], [email protected]
15. Motion strategies for multi-legged robots in unstructured environments Tutors: Darwin Caldwell
Department: Department of Advanced Robotics (ADVR)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description: Legged robots present unique capabilities for traversing a multitude of rough terrains, where
they have the potential to outperform wheeled or tracked systems. The control of such complex robotic
platforms is a challenging task, especially in designs with high levels of kinematic redundancy. A wide range
of approaches to legged locomotion exist; however, most are far from demonstrating a solution which
performs with a level of flexibility, reliability and careful foot placement that would enable practical
locomotion on the variety of rough and intermittent terrain humans negotiate with ease on a regular basis.
The position advertised will focus on bridging this gap in the context of multi-directional navigation of
unstructured environments. The work will be conducted on the IIT quadrupeds HyQ, HyQ2Max and the new
centaur inspired robot HalfMan.
Requirements: background in robotics, computer science, electrical engineering or mechanical engineering.
Understanding of robot kinematics and dynamics, strong C++ skills. Passionate for robotics and legged
locomotion, experienced in ROS and Matlab.
Contacts: [email protected]
16. Robot-Assisted Microsurgery Tutors: Dr. Leonardo Mattos, Dr. Nikhil Deshpande
Department of Advanced Robotics (ADVR – IIT)
http://www.iit.it/en/advr-labs/biomedical-robotics.html
Description: Microsurgeries are demanding operations that required high precision and dexterity. They also
represent a surgical area in which robotics can have a deep impact, helping surgeons perform more precise
and safer operations, or even pioneer previously impossible procedures. This research will contribute to the
area of minimally invasive robot-assisted laser microsurgery. It will build upon results from the European
project µRALP (www.microralp.eu) to create the next generation tools for high precision / high quality laser
microsurgeries. This will involve the mechatronic design and control of a new miniaturized laser
micromanipulator, as well as the evaluation and testing of new systems in collaboration with our partner
surgeons. During this process the student will develop expertise in surgical robotics, medical equipment
design, control systems, user interfaces and usability analysis.
Requirements: background in engineering; interest in the design, fabrication and analysis of robots and
mechanisms for microsurgical applications. Experience in CAD-based mechanical design or microfabrication
are desired. The candidate must be fluent in both spoken and written English.
Contacts: [email protected]; [email protected]
17. Novel Interfaces and Technologies for Assistive Robotic Systems Tutor: Dr. Leonardo Mattos, Dr. Nikhil Deshpande
Department of Advanced Robotics (ADVR – IIT)
http://www.iit.it/en/advr-labs/biomedical-robotics.html
Description: Technology can go a long way toward improving the quality of life of people who happen to
have disabilities, including the elderly and those with debilitating diseases such as amyotrophic lateral
sclerosis (ALS), muscular dystrophy, etc. This PhD program will focus on the creation of novel interfaces and
systems to assist people with disabilities realize fundamental activities such as communication,
environment control, social interactions and the ability to move around independently. It may also involve
the investigate technologies suitable for assisted living using body-area and ambient wireless computing
networks. The research will involve close collaboration with partner clinicians and technology end-users,
allowing the student to develop expertise both in biomedical engineering (biosensors, actuators, control
systems) and ergonomics (human factors, usability, human-computer interaction).
Requirements: background in biomedical engineering, computer science or related disciplines; interest in
the design, implementation and evaluation of assistive systems. Experience in brain-machine interfaces
(BMI) or the acquisition and processing of biosignals would be advantageous. The candidate must be fluent
in both spoken and written English.
Contacts: [email protected]; [email protected]
18. Legged Robots (Humanoids) for Hazardous Environment Interventions Tutors: Darwin G. Caldwell
Department of Advanced Robotics (ADVR – IIT)
http://www.iit.it/advr
Description: The world, both natural and man-made, is a complex, unstructured, cluttered and
dynamically changing environment, in which humans and animals move with consummate ease, avoiding
injury to themselves, damage to the environment and performing simple and complex tasks involving
coordination of arms and legs. Wheeled and tracked robots are increasingly able to work in some of these
terrains, particularly those that have naturally or artificially smoothed and flattened surfaces, but there are,
and ultimately will continue to be, many scenarios where only human/animal-like levels of agility,
compliance, dexterity, robustness, reliability and movement/locomotion will be sufficient.
To operate within infrastructures originally designed for humans, but which are, or have become, hostile or
dangerous, a robot should possess a rich repertoire of human-like skills, and a human or animal inspired
(but not necessarily copied) form. Any robot operating in such conditions should also exhibit physical
power, agility and robustness, manipulation and locomotion capability, and ultimately have the capacity to
reach and physically interact with dials, levers, valves, doors, control surfaces etc within the harsh
environment.
These hazardous human engineered environments create challenges and opportunities which will demand
increased functionality in the legged robots, moving from the current domain dominated by simple walking
and balance maintenance, to address key whole body interaction issues (coordinated upper and lower
body) during physical contact with humans, other robots, and the environment.
This project will explore the use of humanoid robots such as COMAN and WalkMan and WalkMan Jr in
these complex hazardous environments.
Requirements: background in computer science, mathematics, engineering, physics or related disciplines.
Contacts: [email protected]
19. Legged Robots (Quadrupeds) for Hazardous Environment Interventions Tutors: Darwin G. Caldwell
Department of Advanced Robotics (ADVR – IIT)
http://www.iit.it/advr
Description: After Fukushima Daiichi nuclear disaster in 2011 the need for robots to deal with unstructured
environments and replace humans in hazardous tasks became one of the main unsolved problems in
robotics. To operate within infrastructures originally designed for humans, but which are, or have become,
hostile or dangerous, a robot should possess a rich repertoire of human-like skills, and a human or animal
inspired (but not necessarily copied) form. Any robot operating in such conditions should also exhibit
physical power, agility and robustness, manipulation and locomotion capability, and ultimately have the
capacity to reach and physically interact with dials, levers, valves, doors, control surfaces etc within the
harsh environment.
These hazardous human engineered environments create challenges and opportunities which will demand
increased functionality in the legged robots, moving from the current domain dominated by simple walking
and balance maintenance, to address key whole body interaction issues (coordinated upper and lower
body) during physical contact with humans, other robots, and the environment.
This project will explore the use of quadruped robots such as HyQ, HyQ2Max and the centaur inspired
HalfMan to explore robot actions within complex hazardous environments.
Requirements: background in computer science, mathematics, engineering, physics or related disciplines.
Contacts: [email protected]
20. Advanced Teleoperation: Tele-locomotion, tele-manipulation, and human-
machine interaction for extreme environments. Tutor: Darwin Caldwell, Nikhil Deshpande
Positions: 2
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
In some work environments, eg nuclear power plants (explosions, contamination, fire, etc.), Chemical,
petrochemical, steel, and process industries (biological risks , fire, inhalation of smoke or toxic gas, etc.),
construction and demolition waste (work at height/confined spaces or compromised air quality, etc.),
submarine and extractive operations (drowning, lack of oxygen, fumes or toxic gases, etc.), even during
routine activities such as general operation, inspection and maintenance, there can be a significant or
extreme risk to the health and welfare of workers.
Robots and robotics can be an obvious solution to the human risk but although the robot may be able to
physically perform some, most or all of the tasks they lack the cognitive ability to problem solve in complex
environments. However, an advanced, intelligent user interface(s) linked to dexterous, multi-limbed,
walking and wheeled mobile platforms could form a solution to substitute or assist workers in these stages,
reducing or eliminating exposure to hazards.
This project will develop new hardware and master-slave teleoperation software, to operate in high risk
industries with the goal of reducing these risks. The project will use, develop and integrate advanced
technologies on tele-locomotion, tele-manipulation, and human-machine interaction. The project will build
on the strong existing technological capabilities in the Department of Advanced Robotics (ADVR) Italian
Institute of Technology, acquired through the successful implementation of various high-tech projects in
this field. The technical know-how of these previous proje ts, i ludi g H Q, Wea Hap, μRALP, Ro o-Mate,
and Walk-Man, will contribute to the development of an intuitive master-slave teleoperation interface
forming an advanced user interface.
Requirements: The successful candidates will have a Master degree in Mechatronics, Robotics, Engineering
or equivalent and will be able to work both in a team and independently. Experience in mechanical design,
programming with C/C++ and knowledge of robot kinematics and dynamics is preferable. *Note: It is
compulsory to prepare a research proposal on this topic.
Contact: [email protected], [email protected]
21. Intuitive control of industrial exoskeletons Tutor: Darwin Caldwell, Jesus Ortiz
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description: An industrial exoskeleton is a complex wearable robotic device that provides an assistive
force/torque to the operator to help him during a specific task. The assistance requirements might vary
between tasks, operators and/or time. Current industrial exoskeletons use basic control strategies based on
sensory input, pre-set configuration parameters and a few on-line adjustments. However, this is not
enough for a flexible and user-friendly system, also considering the variability between tasks and users. For
this reason, a human-robot collaboration interface including an intuitive command interface and an
adaptive control system is desirable. The research would focus on the study of different intuitive command
interfaces and adaptive control systems, design and development of an intuitive control system, which
combines a novel user interface with a high-level adaptive controller, and testing and evaluation of the
developed systems.
Main contribution: Advanced intuitive control for industrial exoskeletons
Main tasks:
Research on different modalities of intuitive control interfaces
Research on high-level controllers for user intention identification
Research on human motion segmentation
Design and development of a combined command interface and an adaptive high-level controller
for industrial exoskeletons
Testing and evaluation of command interface and controller
Requirements: The successful candidates will have a Master degree in Mechatronics, Robotics, Engineering
or equivalent and will be able to work both in a team and independently. Experience in mechanical design,
programming with C/C++ and knowledge of robot kinematics and dynamics is preferable. *Note: It is
compulsory to prepare a research proposal on this topic.
Contact: [email protected], [email protected]
22. Advanced Physical Human-Robot Interfaces for Exoskeletons Tutor: Darwin Caldwell, Jesus Ortiz
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description: The interaction between the human body and a wearable robot is a critical aspect. The
assistive forces produced by the robot have to be transmitted through the surface of the human body. The
ways the forces are distributed on the human surface determine the effectiveness and comfort of the
exoskeleton. However, the design of the physical interfaces is a challenging task, since the surface of the
human body is not uniform and can vary in time. Furthermore, it is important to consider other aspects
related with the comfort such as temperature, humidity, breathability, biological compatibility, etc. The
research would focus on characterization of the human body surface and modelling of the interaction of
between a wearable robot and the human body, as well as the development and testing of advanced
physical human-robot interfaces for an industrial exoskeleton.
Main contribution: Characterization of the human body for the interaction with a wearable robot
Main tasks:
Research on physical human body interaction
Modelling of the physical interaction between a wearable robot and the human body
Developing a apa it ap of the hu a od
Developing of physical human-robot interfaces for an industrial exoskeleton
Testing and validation of different physical human-robot interfaces
Requirements: The successful candidates will have a Master degree in Mechatronics, Robotics, Engineering
or equivalent and will be able to work both in a team and independently. Experience in mechanical design,
programming with C/C++ and knowledge of robot kinematics and dynamics is preferable. *Note: It is
compulsory to prepare a research proposal on this topic.
Contact: [email protected], [email protected]
23. LEGGED LOCOMOTION-CONTROL: Locomotion Control of a Mobile
Manipulation platform with Hybrid leg and wheel functionality Tutor: Nikos Tsagarakis, Navvab Kashiri
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
Emerging robots operating within man-made real-word workspaces will have to walk, reach, physically
interact, pick up, retrieve and manipulate a variety of objects, tools and interfaces designed for human use.
Such mobile manipulation is an activity that humans naturally perform by combining two motion
capabilities: locomotion and manipulation. This need of mobile manipulation has been tackled in the past
with the development of a variety of mobile manipulation systems made by robotic arms installed on
mobile bases with the mobility provided by wheels and legs mechanisms. On one hand wheeled rovers
provide optimal solutions for well-structured, and relatively flat terrains environments, however, outside of
these types of workspaces and terrains their mobility decreases significantly and usually they can only
overcome obstacles smaller than the size of their wheels. Compared to wheeled robots, legged robots are
more sophisticated to design, build and control but they have obvious mobility advantages when operating
in unstructured terrains and environments.
This research theme will focus on the development of locomotion pattern generators and mobility control
of a CENTAURO form mobile manipulation platform that uses legs and small wheels to combine the
advantages of wheeled and legged locomotion (https://www.centauro-project.eu/). On flat terrains
directly driven wheels will move the robot quickly and efficiently in an omnidirectional way by
independently adjusting their speed and orientation. When driving over uneven ground, the legs will adapt
to the surface, such that the posture of the main body is stabilized. Different principles and combinations of
leg gaits and wheel mobility mechanisms will be developed and evaluated in simulation and finally
implemented and validated on the CENTAURO prototype.
Requirements: We are seeking for highly motivated candidates with a background in Electrical, Control
engineering, Physical Sciences or Robotics. Candidates should have strong competencies in robot dynamics,
control and excellent programming skills in Matlab and C++. (Programing and Simulation 30%, Dynamics
30%, Control %40). The experience on dynamic simulators (e.g. Gazebo, Webot, RoboTran, etc.) and ROS
would be plus. *Note: It is compulsory to prepare a research proposal on this topic.
Reference: N.G.Tsagarakis, S. Morfey, G.Medrano-Ce da, H. Dallali, D.G.Cald ell, A As et i Co plia t A tago isti Joi t Desig fo High Pe fo a e Mo ilit , IEEE I te atio al Co fe e e o Intelligent Robots and Systems (IROS), 2013, pp 5512-5517.
Contact: [email protected]
24. LEGGED LOCOMOTION-CONTROL: Whole-body Sensor Fusion and Motion
Adaptation of a Legged Manipulation Robot Tuto : Nikos Tsaga akis, Na a Kashi ,I, Che g u Zhou.
Depa t e t of Ad a ed Ro oti s Istituto Italia o di Te ologia http:// .iit.it/e / esea h/depa t e ts/ad a ed- o oti s.ht l
Description:
To date, the sensor feedback provided by the sensors equipped in legged robot is very much limited
compared to the rich sensory information exists in humans. Moreover, there is a lack of an effective sensor
fusion frameworks in practice that fuses the different feedback signals, estimates the parameters and the
state of the system which can be subsequently used for the motion control of the machine.
This proposed research theme aims at the development of the whole-body sensory feedback system that
provides a set of fused feedback to assist the locomotion and manipulation control, and monitor the state
of the system. The whole-body sensor framework will be composed by three layers: the first layer is the
elementary signal filtering and automatic offset removal of all types of sensors; the second layer is the
sensor fusion of the processed signals from the first layer, which extract the processed data from the first
layer and compute high level integrated information (for example, the online estimation of the center of
mass of the robot, the kinematics/dynamics state, the identified terrain properties); the last layer will make
an effective use of the extracted information from the second layer to assist and suggest the locomotion
and manipulation control framework about what the status of the robot is, what kind of control/adaptation
mechanisms should be activated or dismissed and so on, as well as managing the system status
report/record for further diagnosis when necessary.
Particularly in the first layer of the whole-body sensor fusion, the candidate is expected to develop self-
calibration algorithms that explore the redundancy of the sensor feedback and an explicit dynamic model
of the robot to calibrate those sensors which have signal drifting issue or temperature varying property.
These algorithms should be able to run offline (a self-calibration procedure as a part of initialization of the
system), and online (while robot is running on the field).
Both the whole-body sensor fusion will serve as a foundation for the control and planning of the
locomotion and manipulation which endow the legged mobile manipulation platform to traverse various
types of terrains. These methods will be experimentally validated using the high performance humanoid
WALK-MAN (www.walk-man.eu) and the quadruped manipulation robot CENTAURO
(https://www.centauro-project.eu/)
Re ui e e ts: Appli a ts should ha e st o g a kg ou d i sig al p o essi g, digital filte desig Kal a filte , et . , a d p og a i g skills i MATLAB a d C/C++. K o ledge o e hat o i s ha d a e, fu da e tal o oti s a d igid od d a i s as ell e pe ie e ith diffe e t si ulatio tools e.g. Gaze o, We ot, Ro oT a , et . is a plus. P og a i g a d Si ulatio %, D a i s %, Co t ol % . *Note: It is o pulso to p epa e a esea h p oposal o this topi . Contact: [email protected]
25. LEGGED LOCOMOTION-CONTROL: Robust Locomotion of Humanoids in an
Unstructured Environment Under External Forces Tutor: Chengxu Zou, Nikos Tsagarakis
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
Recent developments, especially in relation to DARPA challenge, brought numerous improvements to the
capabilities of humanoid robots. There are however still problems that need to be solved before we can
apply humanoids to practical tasks. One of them is a robust locomotion of humanoids in an unknown
environment under external forces and disturbances. Whatever practical task we will want a humanoid
robot to perform, it will involve physical interaction with an environment and/or objects. Also we cannot
assume that the robots will have a comfort of walking on a flat and structured ground. This brings us to the
topic of this research. Robust locomotion of a humanoid robot on an uneven terrain subject to forces
coming either from objects being manipulated by humanoid or from obstacles.
The successful candidate will work on real-time locomotion control. This will involve gait and whole-body
motion planning, given the perceived map of environment; whole body state estimation based on
proprioceptive and exteroceptive sensory data; stabilization control and gait re-planning based on the
perceived information. All algorithms developed in this work should be implemented and tested on our
new humanoid robot WALK-MAN. It has been developed under the European FP7 project WALK-MAN
(http://www.walk-man.eu/). WALK-MAN is an adult size humanoid robot designed with an objective of
application in search and rescue scenarios. The robot has compliant joint structures, 6 axis Force/Torque
sensors at the ankles and the feet soles are equipped with pressure sensing arrays . These, together with
MultiSense head allow for proprioceptive and exteroceptive sensing.
Requirements: Applicant should possess strong background in: physical system, modeling and rigid body
dynamics, robust control theory, MATLAB and C/C++ programing languages. Previous experience with
biped locomotion is a plus.The following qualities will be advantageous: knowledge of optimization
techniques (QP, SQP etc.), knowledge of signal processing (KF, EKF etc.), hands-on experience with robotic
platforms. *Note: It is compulsory to prepare a research proposal on this topic.
Reference: P ze sla K zka, Peta Ko ushe , Nikos G Tsaga akis, Da i G Cald ell, O li e ege e atio of ipedal alki g gait patte opti izi g footstep pla e e t a d ti i g , Intelligent Robots
and Systems (IROS), 2015 IEEE/RSJ International Conference on, pp3352-3357, 2015.
Contact: [email protected] , [email protected]
26. LEGGED LOCOMOTION-CONTROL: Dexterous Humanoid Walking on
Restricted and Unstable Footholds Tutor: Nikos Tsagarakis, Chengxu Zou
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
Despite the significant progress made in Humanoid locomotion during the last decade most of the present-
day humanoids still suffer from major problems related to stable walking in terrains other than even. Flat
terrains are though very ideal surfaces compared to terrains existing in human environments where stairs,
inclined surfaces, small obstacles and even rough surfaces may exist. Up to now, there are only few
effective demonstrations about walking and motion planning on this kind of environments. A new
humanoid robot WALK-MAN has been developed under the European FP7 project WALK-MAN
(http://www.walk-man.eu/).). This newly developed robot has compliant joint structures, 6 axis
Force/Torque sensors at the ankles and the feet soles are equipped with pressure sensing arrays which
permit to explore walking on:
a) Uneven terrains and stepping on obstacles of limited surface that may result in limited area footholds.
b) Particulate solid surfaces consisting of particles of different size and density which may not provide
fully stable footholds.
In this project techniques will be developed to plan the execution of dexterous foot probing and regulate
the gait motions accordingly ensuring both the dynamic equilibrium and body/feet posture of the
humanoid to achieve walking on uneven surfaces of limited support area avoiding or stepping on obstacles
with variable inclinations, on unstable particulate surfaces such as terrains composed of small stones.
These methods will take into account kinematics/dynamics and self-collision constraints while detection of
the terrain properties will be assisted by rich sensory feedback from the feet of the humanoid. We will
explore how to detect rough terrain/obstacle properties such as size, inclination and stability using the
sensorized ankle and feet of the humanoid. Having determined the rough terrain characteristics, how the
balance stability will be maintained when the robot will be on this specific rough terrain will be evaluated
and different control and trajectory planning methodologies will be developed to allow the humanoid to
pass through while maintaining stability and balance.
Requirements: Applicant should ideally possess strong background in physical system modeling and
control, MATLAB and C/C++ programming. Knowledge on mechatronics hardware, fundamental robotics
and rigid body dynamics is a plus. *Note: It is compulsory to prepare a research proposal on this topic.
Reference: Che g u Zhou, Xi Wa g, Zhi i Li, Da i Cald ell, Nikos Tsaga akis, E ploiti g the redundancy for humanoid robots to d a i all step o e a la ge o sta le , I tellige t Ro ots a d S ste s (IROS), 2015 IEEE/RSJ International Conference, pp 1599-1604, 2015
Contact: [email protected]
27. CONTROL: Human-Robot Collaborative Control of a Mobile Manipulation
platform with Hybrid leg and wheel functionality Tutors: Nikos Tsagarakis, Arash Ajoudani,
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
The robots are coming out of the cage, and getting closely involved into human life and physically
interacting with them to execute tasks in a collaborative manner.
This research theme will focus on the development of a collaboration control framework that will permit a
human operator to perform heavy manipulation tasks in collaboration and direct interaction with a
CENTAURO form mobile manipulation platform that uses legs and small wheels to combine the advantages
of wheeled and legged locomotion (https://www.centauro-project.eu/). The project will look on the
development of human motion/impedance intention estimation modules and collaboration controllers that
take into account the human motion and impedance estimations to command and drive the execution of
the manipulation task. Autonomous motion and impedance regulation principles will also be applied at the
robot side to assist the human partner in commanding the collaborative behaviours of the robot assistan.t.
Requirements: We are seeking for highly motivated candidates with a background in Electrical, Control
engineering, Physical Sciences or Robotics. Candidates should have strong competencies in robot dynamics,
control and excellent programming skills in Matlab and C++. (Programing and Simulation 30%, Dynamics
30%, Control %40). The experience on dynamic simulators (e.g. Gazebo, Webot, etc.) and ROS would be
plus. *Note: It is compulsory to prepare a research proposal on this topic.
Reference: L. Pete el, N.G. Tsaga akis, D.G. Cald ell, Da i a d A. Ajouda i , Adaptatio of o ot physical behaviour to human fatigue in human-robot co- a ipulatio , IEEE-RAS 16th International
Conference on Humanoids, pp489-494, 2017
Contact: [email protected]
28. CONTROL: Whole-body Control of Hybrid Humanoid-Quadruped Robotic
Platform Tutor: Jinoh Lee, Nikos Tsagarakis
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
Recent advances in humanoid research aim at bringing robots into the real world which is complex,
unstructured, cluttered and dynamically changing and encourage to operate in various scenarios where
only human-like levels of agility, dexterity, robustness, reliability and manipulation/locomotion will be
sufficient. This research will cover these issues in hybrid robotic system with legged loco-manipulation
using a high performance and high efficiency electric platform, that will combine a quadruped legged base
with high dexterity upper body with dual arms – inspired by the Greek mythology, Centaurs. This
combination will amalgamate the talents and capabilities of highly dynamic quadrupedal locomotion over
rough and unstructured terrain with a humanoid inspired upper body and arms to provide a high dexterity
bimanual manipulative structure.
This research therefore will focus on exploiting an advanced whole-body control strategy of hybrid robotic
platform to guarantee agile responses and high dexterity to cope with multi contacts of dual arm and four
legs and the high kinematic redundancy. It also will include the high-level motion planning to provide
excellent manipulation skills in certain tasks with multiple sensors. The methods will be verified in dynamic
simulation environment such as Gazebo, Robotran, and V-REP; and will be demonstrated on physical robots
such as Centaur robot which is being developed under the European H2020 project CENTAUR
(http://www.centaur-project.eu/ ) and PHOLUS project. This project topic is multidisciplinary, thus the
collaboration with other members will be encouraged, e.g. locomotion and 3D perception.
Requirements:
We are preferably seeking for highly motivated candidates with a background in robotics and control
engineering. Especially, knowledge on robust control theory and operational-space control with redundant
robots will accelerate the progress on this phD theme. This is a multidisciplinary project where the
successful candidates should have strong competencies in robot kinematics/dynamics/control and in
software coding (e.g. C++ in Linux and MATLAB). The experience on dynamic simulators (e.g. Gazebo,
RoboTran, V-REP) is mandatory and ROS would be plus. *Note: It is compulsory to prepare a research
proposal on this topic. *Note: It is compulsory to prepare a research proposal on this topic.
Reference: Jinoh Lee, Houman Dallali, Maolin Jin, Darwin G. Caldwell, Nikos Tsagarakis' "Robust and
Adaptive Whole-body Controller for Humanoids with Multiple Tasks under Uncertain Disturbances," 2016
IEEE International Conference on Robotics and Automation (ICRA 2016), Stockholm, Sweden, May, 2016,
Jinoh Lee, Pyung Hun Chang, Rodrigo S. Jamisola Jr., "Relative Impedance Control for Dual-Arm Robots
Performing Asymmetric Bimanual Tasks," IEEE Trans. on Industrial Electronics, Vol. 61, No. 7, July 2014
Contact: [email protected], [email protected]
29. ROBOTIC ACTUATION: New Efficient actuation systems based on the
blending of Forced and Natural dynamics Tutor: Nikos Tsagarakis, Jorn Mazhan
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
The department of Advances Robotics is currently one of the world leading research institutes on the
development and new robotic actuation systems ranging from series compliant actuators to actuators with
variable compliance and damping characteristics. Elastic components in the actuation may improve the
motion efficiency of the robotic system through energy storage and release during locomotion or permit to
generate high power motions during throwing, kicking and jumping actions. However, this energy
efficiency improvement has not yet demonstrated in real robotic systems. This research theme will explore
the development of a new actuator idea that permits a joint to switch from natural to forced dynamics
through novel transmission systems. Different principles and implementations of this actuation will be
evaluated in simulation and finally implemented and validated on single joint proof of concept prototypes.
The mechanical design developments will accompanied by activities on the regulation strategies of these
new actuation systems to maximize their efficient operation. This is a high risk research theme on
innovative actuation systems which can potentially generate high pay off in terms of novel outcome and
dissemination of results.
Requirements: We are seeking for highly motivated candidates with a background in Electronic/Mechanical
engineering, Physical Sciences or Robotics. Candidates should have competencies in CAD mechanical design
and/or robot dynamics and control. (Mechanical design 50%, Simulation/Dynamics/Control %50). *Note: It
is compulsory to prepare a research proposal on this topic.
Reference: N.G.Tsagarakis, S. Morfey, G.Medrano-Ce da, H. Dallali, D.G.Cald ell, A As et i Compliant Antagonistic Joint Design for High Performance Mo ilit , IEEE I te atio al Co fe e e o Intelligent Robots and Systems (IROS), 2013, pp 5512-5517.
Contact: [email protected]
30. ROBOTIC ACTUATION: Design principles and control of high performance
robotic actuation systems Tutor: Nikos Tsagarakis, Navvab Kashiri
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
The department of Advances Robotics is currently one of the world leading research institutes in the
development and new actuation systems ranging from series compliant actuators to actuators with variable
compliance and damping characteristics. Elastic components in the actuation may improve the motion
efficiency of the robotic system through energy storage and release during locomotion or permit to
generate high power motions during throwing, kicking and jumping actions. However, this energy
efficiency improvement has not yet demonstrated in real systems powered by compliance actuators.
Recently a new actuation concept based on an asymmetric antagonistic scheme has been developed at the
Department of Advanced Robotics. This research will investigate this novel joint actuation and elastic
transmission system and their associate control schemes and eventually demonstrate energy efficient and
high peak operation using large energy storage capacity elements, efficient actuation drivers and energy
recycling techniques. The developed joint concepts and controllers will be eventually applied to walking,
hopping and in general legged robots. The research in the utility and control of the actuator will be also
applied in high power bursts such as throwing, kicking and jumping of anthropomorphic robots. The work
activity of this theme will be in line with the workplan of the WALK-MAN EU project (http://www.walk-
man.eu/)
Requirements: We are seeking for highly motivated candidates with a background in Electronic/Mechanical
engineering, Physical Sciences or Robotics. Candidates should have competencies in CAD mechanical design
and/or robot dynamics and control. (Mechanical design 50%, Dynamics/Control %50). *Note: It is
compulsory to prepare a research proposal on this topic.
Reference: N.G.Tsagarakis, S. Morfey, G.Medrano-Ce da, H. Dallali, D.G.Cald ell, A As et i Compliant Antagonistic Joint Desig fo High Pe fo a e Mo ilit , IEEE I te atio al Co fe e e o Intelligent Robots and Systems (IROS), 2013, pp 5512-5517.
Contact: [email protected]
31. ROBOTIC ACTUATION: Design and Control of Novel Lightweight Robotic
Joint-Link Modules with distributed variable stiffness Tutor: Jörn Malzahn, Nikos Tsagarakis
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
The department of Advances Robotics is currently one of the world leading research institutes in the
development and new actuation systems ranging from series compliant actuators to actuators with variable
compliance and damping characteristics. The intrinsic compliance absorbs shocks during accidental impact
on the joint level. This equally contributes to the protection of gears, sensors as well as objects and humans
in the vicinity of the robot and enables the sensing of interactive contacts. In addition, elastic components
in the actuation may improve the motion efficiency of the robotic system through energy storage and
release during locomotion or permit to generate high power motions during throwing, kicking and jumping
actions.
Despite the recent advances in this field, robotic actuation still does not match the efficiency and power
capacity observed animals and humans. This severely limits the capabilities and operation times of today's
robotic systems e. g. in robotic assistance or disaster response scenarios.
The objective behind this topic is to devise novel robotic joint-link modules, which maximize the amount of
storable energy in relation to the overall mechanism weight in comparison. Therefore, this work will
integrate the principle of variable compliance on both, the actuator as well as the link level along with
suitable control techniques. This holistic and integrated approach to variable elastic actuation is envisioned
to increase the design degrees of freedom of future high performance robot mechanisms.
Requirements: We are seeking for highly motivated candidates with a background in Mechanical
engineering, Control Theory or Robotics. This is a multidisciplinary topic where the successful candidates
should have experiences in CAD mechanism design and strong competences in robot kinematics, dynamics
and control. (Mechanical design 30 %, Kinematics/Dynamics/Control 70 %). *Note: It is compulsory to
prepare a research proposal on this topic.
Reference:
N.G. Tsagarakis, Stephen Morfey, Gustavo Medrano-Cerda, Zhibin Li, Da i G. Cald ell, Co plia t Hu a oid COMAN: Opti al Joi t Stiff ess Tu i g fo Modal F e ue Co t ol , IEEE I te atio al Conference on Robotics and Automation, ICRA 2013, pp 673-678.
J. Malzah , R. F. Rei ha t, T. Be t a D a i s Ide tifi atio of a Damped Multi Elastic Link Robot Arm
u de G a it , IEEE I te atio al Co fe e e o Ro oti s a d Auto atio , ICRA , pp -2175.
Contact: [email protected], [email protected]
32. MULTIMODAL PERCEPTION: 3D Perception for Rough terrain locomotion
and Free-form object manipulation Tuto : Di it is Ka oulas, Nikos Tsaga akis
Depa t e t of Ad a ed Ro oti s Istituto Italia o di Te ologia http:// .iit.it/e / esea h/depa t e ts/ad a ed- o oti s.ht l
Description:
Afte Fukushi a Daii hi u lea disaste i the eed of o ots to deal ith u st u tu ed e i o e ts a d epla e hu a s i haza dous tasks e a e o e of the ai ope p o le s i o oti s. Rapid ad a e e ts i a tuatio a d o t ol o e the last fe ea s e a led a ti ulated hu a oid o ots to oth
alk i u e e te ai a d pe fo de te ous a ipulatio usi g thei ha ds. These apa ilities a e usuall gai ed ithout usi g D pe eptio at all, assu i g that eithe the e i o e t is ostl k o a d
ell-st u tu ed, o the u e tai t a e tole ated lo -le el feed a k o t ol. I eal o ld s e a ios these assu ptio s a ot hold. D pe eptio is e ui ed! The p o le of foot pla e e t i ough te ai fo e a ple i a o k t ail fo alki g o the p o le of g aspi g f ee-fo ed o je ts fo e a ple a o k
usi g D pe eptio e ai s o e of the e t al halle ges i o oti s a d is the ke aspe t fo o pleti g lo o otio o a ipulatio tasks i u k o e i o e ts. The ai of this topi is to de elop e e i o e t e o st u tio te h i ues that e a le hu a oid o ots to pe fo oth legged lo o otio a d a ipulatio tasks i u st u tu ed e i o e ts usi g 3D pe eptio fo foot o ha d pla e e t. The state-of-the-a t D pe eptio se so s ill e used ste eo/ti e-of-flight a e as, lase se so s, o st u tu ed light s ste s alo g ith othe pe eptio se so s like ta tile, fo e o t ol, o IMU o es. The de se D poi t loud that is a ui ed f o a a ge se so ill e ui e so e geo et i si plifi atio s fo easo i g the o ta t et ee the o ot's foot/ha d a d a a ea i the e i o e t. Modeli g these o ta t a eas a ou d a d o a o ot hile usi g Si ulta eous Lo alizatio a d Mappi g SLAM
te h i ues fo eati g a d keepi g a ap of these ith espe t to the o ot is a ke aspe t fo o pleti g these tasks. The de eloped ethods ill e tested oth i si ulatio a d o a eal full-size hu a oid o ot WALK-MAN . The p oje t is i te dis ipli a si e pe eptio eeds to e o i ed ith path pla i g
a d o t ol te h i ues fo aki g the a tual o ot o plete a task. Thus the olla o atio ith othe e e s of the p oje t ill e e ui ed i to that di e tio . The o k a ti it of this the e ill e i li e ith the de elop e ts of the WALK-MAN EU p oje t http:// . alk- a .eu/ .
Re ui e e ts: This topi lies i the i te se tio of Co pute Visio a d Ro oti s. Ideal appli a ts should ha e st o g a al ti al a d p og a i g skills C/C++ a d MATLAB . A ele a t deg ee is e ui ed, fo i sta e i Co pute S ie e o E gi ee i g. A a kg ou d i Ro oti s/Co pute Visio is desi a le, hile k o ledge of the Ro ot Ope ati g S ste ROS , the Poi t Cloud Li a PCL , o the Ope Sou e Co pute Visio Li a Ope CV is a ig plus. The appli a ts should e flue t i E glish a d tea pla e s. *Note: It is o pulso to p epa e a esea h p oposal o this topi . Reference: Dimitrios Kanoulas, "Curved Surface Patches for Rough Terrain Perception", Ph.D
Thesis, 2014.
Contact: [email protected], [email protected]
33. MULTIMODAL PERCEPTION: Cutaneous and Kinesthetic sensing for robotic
arms, dextrous hands and feet Tuto : Ioa is Sa akoglou, Nikos Tsaga akis
Depa t e t of Ad a ed Ro oti s Istituto Italia o di Te ologia http:// .iit.it/e / esea h/depa t e ts/ad a ed- o oti s.ht l
Description:
Tactile/force sensing is an important area in haptics, teleoperation and robotic dexterous manipulation.
Manipulation of objects through robotic hands can only be efficiently performed if the interaction between
the robotic hand and the object is effectively sensed. Similarly precise and careful locomotion requires the
sense of interaction between the foot and the ground that goes beyond the standard F/T sensing modules
usually integrated in the feet of legged robots. Currently research and development in tactile/force sensing
is directed toward anthropomorphic sensors which attempt to match the sensing capabilities of the human
skin and which resemble its mechanical properties. This proves to be a great and exiting challenge. This
research will focus on new tactile/force sensing technologies suitable for application on robotic arms,
hands and feet. It will involve research in the areas of distributed pressure tactile sensing in the form of
highly anthropomorphic/bio-mimetic artificial skins and force sensing with semi-rigid skins in the form of
high accuracy monolithic force/torque sensors. New sensor designs will be sought based on the current
sensing technologies such as resistive, capacitive, piezoelectric, piezoresistive or new materials such as
nano-particle filled polymers. Appropriate methods will be sought for developing and integrating large
populations of sensing elements into structures suitable to operate as robotic skins. The candidate will also
tackle the technological challenges in connectivity, power, and the signal processing of the distributed
sensor. The candidate will work closely within a team of researchers and technicians toward developing
working systems with a final goal to integrate tactile sensing in humanoid platforms such as COMAN
(http://www.iit.it/en/advr-labs/humanoids-a-human-centred-mechatronics/advr-humanoids-
projects/compliant-humanoid-platform-coman.html) and WALK-MAN (http://www.walk-man.eu/).
Requirements: The ideal candidate will be a talented individual with an Electronics or Mechatronics
background and a strong performance in hardware design projects. The candidate should be willing to work
in diverse areas, ranging from simulation (MATLAB, Maple Sim, etc), hardware design and software
development (C++). *Note: It is compulsory to prepare a research proposal on this topic.
Reference: I. Sarakoglou, N. Garcia-Hernandez, N. G. Tsagarakis, and D. G. Caldwell, "A High Performance
Tactile Feedback Display and Its Integration in Teleoperation," IEEE Transactions on Haptics, vol. 5, no. 3,
pp. 252-263, 2012.
Contact: [email protected], [email protected]
34. WEARABLE ROBOTICS: Development of Under-actuated upper limp
wearable systems for Teleoperation and rehabilitation Tutor: Ioannis Sarakoglou, Nikos Tsagarakis
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
This theme focuses on the development of wearable kinesthetic devices for the upper limp including
devices for the hand and arm. One of the main objectives of the design of these systems is to move away
from the traditional design paradigms of upper limp exoskeleton devices that target to develop systems
with many actuators following anthropomorphic exoskeleton structures attached to the upper limp
segments using physical interfaces with multiple fixation points. With special attention on the systems
ergonomics both a t the level of physical interface as well as at the level of the functionality this project will
follow instead a different approach that will permit the development of devices which can execute closely
functional rehabilitation as that provided by physiotherapist towards systems that combine the benefits of
robotic aid automated rehabilitation with the natural execution of physiotherapy regimes as performed by
the dedicated medical personnel. At the implementation and physical interface level the project will
attempt to minimize the complexity yet keeping the functionality of the device through the use of under-
actuation and the employment of minimalistic physical interface principles that can resemble the
i te a tio et ee the ph siothe apist a d the patie ts’ uppe li p. The a ti it of this p oje t is strongly linked to the recently started EU H2020 project SOFTPRO (http://www.softpro.eu/)
Requirements: The successful candidates will have a Master degree in Mechatronics, Robotics, Mechanical
Engineering or equivalent and will be able to work both in a team and independently. Experience in CAD
mechanical design, programming with C/C++ and Matlab is mandatory and knowledge of robot kinematics
and dynamics is preferable. (60% mechanical design, 20% control, 20% software). *Note: It is compulsory to
prepare a research proposal on this topic.
Reference: J I al, H Kha , NG Tsaga akis, DG Cald ell, A o el e oskeleton robotic system for hand
rehabilitation–Co eptualizatio to p otot pi g , Bio e eti s a d io edi al e gi ee i g , 9-89
Contact: [email protected], [email protected]
35. WEARABLE ROBOTICS: Development of EXO-Muscles: Wearable Add-On
Elastic power force augmentation Units Tutor: Nikos Tsagarakis, Ioannis Sarakoglou, Arash Ajoudani
Department of Advanced Robotics (Istituto Italiano di Tecnologia)
http://www.iit.it/en/research/departments/advanced-robotics.html
Description:
This project targets on the development of power autonomous, intelligent single muscle-type actuation
units to act as power/force augmentation devices of individual joints of the human limbs (arms or legs). The
term "wearable" implies for a portable, lightweight systems favouring comfort and ergonomics. The
improvement of the wearability of the EXO-Muscles will be therefore considered during the development
process and optimizations will be applied in all stages of the mechatronic developments related to the
actuation system, the device structure and physical interface to the human limb. In contrast to the multidof
highly complex force reflecting robotic exoskeletal structures, this unit can form the primitive force
augmentation unit for building wearable force feedback systems with improved ergonomics and comfort.
We envisage the development of 1 or 2 DOF systems e.g. an elbow device, or elbow/wrist or a knee/hip
system. The regulation of the assistive forces will be performed considering control schemes built around
rich sensing state feedback that will include traditional force/torque sensing technologies in conjunction
with biofeedback modalities that will allow the estimation of human effort and joint fatigue. An additional
rich sensory interface will allow the estimation of the human body posture, motion intention/measurement
a d hu a /e i o e t o ta t state. Based o this the assisti e ope atio ill e i tellige tl tu ed to
ensure that the appropriate level of assistance is delivered. The activity of this project is strongly linked to
the recently started EU H2020 project SOFTPRO (http://www.softpro.eu/)
Requirements: The successful candidates will have a Master degree in Mechatronics, Robotics, Mechanical
Engineering or equivalent and will be able to work both in a team and independently. Experience in CAD
mechanical design, programming with C/C++ and Matlab is mandatory and knowledge of robot kinematics
and dynamics is preferable. (60% mechanical design, 20% control, 20% software). *Note: It is compulsory to
prepare a research proposal on this topic.
Reference: Nikos Karavas, Arash Ajoudani, N. G. Tsagarakis, Jody Saglia, Antonio Bicchi, Darwin G. Caldwell,
Tele-i peda e ased Assisti e Co t ol fo a Co plia t K ee E oskeleto , Ro oti s a d Auto o ous Systems 2014.
Contact: [email protected], [email protected]
36. Locomotion Planning and Adaptation Strategies for Multi-Legged Robotic
Platforms on Soft Terrains Tutors: A d eea Rădules u, Vi to Ba asuol, Claudio Se i i
Research Line: Dynamic Legged Systems lab, Dept. of Advanced Robotics (IIT)
http://www.iit.it/hyq
Description:
The Hydraulic Quadruped robot - HyQ - is a fully torque-controlled hydraulically actuated quadruped robot
[link], capable of locomotion over rough terrain and performing highly dynamic tasks such as jumping and
running with a variety of gaits. It is a unique research platform, designed for unstructured environments,
e.g. outdoors and disaster sites.
Legged robots present unique capabilities for traversing challenging terrains, where they offer a clear
advantage over wheeled platforms. A wide range of approaches to legged locomotion exist; however, most
are far from demonstrating a solution which performs with a level of flexibility, reliability and careful foot
placement that would enable practical locomotion on the variety of rough, soft or intermittent terrains. The
success of these behaviours often relies on tuning a large number of parameters, which can be a lengthy
empirical procedure, requiring an expert user.
This position will focus on bridging this gap in the context of locomotion adaptation. The study will focus on
reshaping an initial whole body trajectory (optimal for stiff terrain surfaces). By applying machine learning
methods we can allow the adaptation of this trajectory to perform locomotion on softer surfaces, e.g.,
sand, pebbles and mud. This will allow the robot to robustly navigate a much wider range of natural terrain
scenarios. The work will be conducted on the HyQ and HyQ2Max robotic platforms.
Ideal requirements: Background in robotics, computer science, electrical or mechanical engineering.
Understanding of robot kinematics and dynamics, strong C++ skills. Highly-motivated and passionate for
robotics and legged locomotion, experienced in Matlab and/or ROS. Previous knowledge of machine
learning techniques is a plus.
Contacts: [email protected], [email protected], [email protected]
37. Force Harvesting with Short-Term Contacts for Multi-Legged Robot
Locomotion on Movable and Slippery Terrain Tutors: Michele Focchi, Victor Barasuol, Claudio Semini
Research Line: Dynamic Legged Systems lab, Dept. of Advanced Robotics (IIT)
http://www.iit.it/hyq
Description:
In real life scenarios a legged robot is facing an environment which is continuously changing. From the
locomotion point of view this means that the foot-to-ground contact stability is not ensured along the leg
support period. Indeed, the interaction with the terrain can have its own dynamics (e.g. moving rocks and
slippery surfaces).
A possible strategy to perform locomotion on such challenging surfaces is to understand this dynamics,
exploiting it to generate (up to some extent) proper ground reaction forces to drive the robot. To tackle this
problem we will investigate the concept of "force harvesting" where we design suboptimal contact forces
considering that the interaction will be limited in time (e.g. when stepping on rolling rocks or slippery
surfaces). The research will be conducted and experimentally evaluated on our quadruped robotic
platforms HyQ and HyQ2Max.
Ideal requirements: Background in robotics, computer science, electrical engineering or mechanical
engineering. Understanding of robot kinematics and dynamics, strong C++ skills. Passionate for robotics and
legged locomotion, experienced in ROS and Matlab.
Contacts: [email protected], [email protected], [email protected]
38. Flexible Systems Development for Industrial Applications Tutors: Ferdinando Cannella, Carlo Canali, Francesco Nori, Arash Ajoudani and Darwin Caldwell
ADVANCED INDUSTRIAL AUTOMATION LAB
http://www.iit.it/en/advr-labs/advanced-industrial-automation.html
ADVANCED ROBOTICS (Italian Institute of Technology)
https://www.iit.it/en/research/departments/advanced-robotics.html
Description:
In these years, we are in the midst of a fourth wave of technological advancement: the rise of new digital
industrial technology known as Industry 4.0, a transformation that makes possible to gather and analyse
data across machines, enabling faster, more flexible and more efficient processes to produce higher-quality
goods at reduced costs. This new era of digital enterprise is based on nine technology advances [1], in
particular on data collecting from humans and machines for robotic numerical modelling and simulation
(for machine design, for product development, for manufacturing process, for value chain management,
etc.) and on autonomous robots (to improve both the flexibility and the machine-human collaboration . It’s matter of fact that two of the ultimate basic ideas of the Industrial 4.0 are:
1) the Digital Twin, that is an exact copy (numerical model) of the physical mock-up and that permits
to the machine designers to exploit the power of digitalization to achieve improved efficiency and
quality;
2) the collaborative robot that should interact with worker in the process lines, both from safety point
of you and, at same time, dealing with more complex and heavy task (i.e. sawing, hammering,
drilling, milling, etc.) that requires high compliance and impact energy dumping.
Nevertheless, the current industrial robots are still not flexible (i.e. to high mix low volume manufacturing)
nor adaptive (to the machine-human collaboration) yet.
In this scenario a new robotic design approach should be developed in order to quickly deploy both the
robotic flexibility and adaptivity into robotic manufacturing systems (which may include also multiple arm
or robots, multiple sensors and related machines).
Then the goals of this PhD will be:
study, design and build a novel flexible and adaptive manipulator based on the experience
obtained from previous projects (Autorecon, AvioAero, EuroC, etc.) inspired from the nature of
human beings hands, arms and/or other similar bio-mechanisms.
develop a Digital Twin improving the current virtual prototyping development (co-simulation with
control and flexible multi-body)
include software level based on ROS (or ROS-industrial) which is modular, reconfigurable, adaptive,
easy to use to integrate and control various robotic systems will be developed. The study include
the use of such a software to integrate data coming from sensors, other devices, robots and to
integrate the system in a network (IOT).
The research is carried out within the Advanced Industrial Automation Lab (AIAL is the lab within
Advanced Robotics Department where the innovative, multidisciplinary approach to humanoid design and
control, and the development of novel robotic components and technologies are addressed to the
industrial needs) and in strictly collaboration with the Dynamic Interaction Control (the DIC activities aim
at endowing humanoids with advanced action and physical interaction capabilities), the Human-Robot
Interfaces and Physical Interaction H I ai s to…. , the Center for Micro-BioRobotics (CMBR) and the
MSC software (undisputed worldwide leader for multibody simulations). This encompasses activities from
both the hard and soft systems areas of robotics. Thus the industrial developments exploit these advances
that permit to design the bio-inspired robots suitable for the industrial plants.
[1]https://www.bcgperspectives.com/content/articles/engineered_products_project_business_industry_4
0_future_productivity_growth_manufacturing_industries/
This multidisciplinary work will be under supervision of Dr. Ferdinando Cannella (AIAL), Dr. Mariapaola
D’I pe io AIAL , D . Ca lo Ca ali AIAL , Dr. Francesco Nori (DIC), Dr. Arash Ajoudani (HRI2), Dr. Barbara
Mazzolai (CMBR) and Dr. Luca Castignani (MSC.Software)
Requirements:
this position is open to a PhD candidate should be enthusiastic, with strong interesting in bio-inspired
mechanism and skill in computer science; moreover the robotic modelling will be valuable. The ideal
competencies should be in control with strong programing skill, especially program using C++ under Linux.
The background must have also robotics with skill in using of flexible multi-body simulators (i.e.
MSC.ADAMS, NASTRAN, MARC, etc.).
Required technical skills: 70% control, 30% Mechanism.
Reference:
Ma iapaola D’I pe io, Lu a Ca o a i, Alsa egh, M., Be t a , T., Matteo Palpa elli, Da i G. Cald ell,, Ferdinando Cannella, Fle i le o oti a d a i s th ough defle tio s e app oa h MESA 2016 - 12th IEEE/ASME International Conference on Mechatronic and Embedded Systems and Applications
- Conference Proceedings, art. no. 7587169, DOI: 10.1109/MESA.2016.7587169
L. Fiorio, F. Romano, A. Parmiggiani, G. Sandini, and Francesco Nori. Stiction compensation in agonist-
antagonist variable stiffness actuators. In Proceedings of Robotics: Science and Systems, Berkeley, USA, July
2014.
A. Ajoudani, N. Tsagarakis, and A. Bicchi, "Choosing Poses for Force and Stiffness Control", 2016, submitted.
B Mazzolai, V Mattoli, Robotics: Generation soft, Nature 536 (7617), 400-401, 2016.
http://www.maxwellsci.com/print/rjaset/v6-3778-3783.pdf
Contacts: [email protected], [email protected], [email protected], [email protected]
39. Industrial Robotics for Inspection and Maintenance Tutors: Carlo Canali, Ferdinando Cannella, Darwin Caldwell
ADVANCED INDUSTRIAL AUTOMATION LAB
http://www.iit.it/en/advr-labs/advanced-industrial-automation.html
ADVANCED ROBOTICS (Italian Institute of Technology)
https://www.iit.it/en/research/departments/advanced-robotics.html
Description:
The successful PhD candidate will be involved into the design and study of inspection robots to be used in
industrial applications. The goal of the study is to develop one or more mechatronics systems to be used as
a versatile and flexible device for inspection porpoise. The study of the state of the art in robotics referred
to the current status of manufacturing industry is part of the thesis. The analysis of most demanding
needing in industrial applications will drive the development of the device. The robot to be designed
integrates several sub-systems and can be deployed in several tasks ranging from inspection, maintenance
and quality control. The system must be able to access small spaces, navigate in harsh environments and
carry on a number of sensors and actuators depending from the application.
As a general rule the system need to be Industry 4.0 complaints including the following features:
Inter-operability: The ability of machines, devices, sensors, and people to connect and communicate with
each other, aggregation of raw sensor data to higher-value context information.
Physically support humans by conducting a range of tasks that are unpleasant, too exhausting, or unsafe
for their human co-workers. This could include telepresence, remote operation, and augmented reality.
The ability of the systems to make decisions on their own and to perform their tasks as autonomously as
possible.
This work will be under supervision of Dr. Carlo Canali and Dr. Ferdinando Cannella
Requirements:
Background in mechatronics, mechanical engineering, electronic engineering, or equivalent
Desired qualifications:
The successful applicant is expected to have a strong background in electronics, mechanics or mechatronics
Experience with real hardware, and excellent hands-on practical skills are essentials, the ideal candidate
has a strong proactive attitude and problem solving capabilities.
Team working and competence in one or more of the following topics are required:
Mechatronics
Mechanical or Electronic design
Control and software engineering
Reference:
Carlo Canali, Ferdinando Cannella, Fei Chen, Traveler Hauptman, Giuseppe Sofia, Amit A. Eytan, Darwin G.
Caldwell "High Reconfigurable Self-Adaptive Robotic Gripper for Flexible Assembly" in Proceedings of the
ASME 2014 International Design and Engineering Technical Conferences & Computers and Information in
Engineering Conference, IDETC/CIE 2014, August 17-20, 2014, Buffalo, NY
Fe di a do Ca ella, Al e to Ga i ei, Ma iapaola D’I pe io a d Gia lu a Rossi, A No el Method Fo The Desig Of P ostheses Based O The oelasti St ess A al sis A d Fi ite Ele e t A al sis Jou al Of Mechanics In Medicine And Biology Vol. 14, No. 5 (2014) 1450064, World Scientific Publishing Company,
Doi: 10.1142/S021951941450064.
Ma iapaola D’I pe io ,Fe di a do Ca ella, Fei Che , Da iele Catela i, Claudio Se i i a d Da i G. Cald ell Modelli g Legged Ro ot Multi-Body D a i s Usi g Hie a hi al Vi tual P otot pe Desig -Proceedings of Living Machines'14 Proceedings of the Second international conference on Biomimetic and
Biohybrid Systems.
Contacts: [email protected], [email protected]